With AP2, AI agents can securely complete purchases—whether by card or crypto—while providing users with transparent, verifiable proof of transactions. Image Source: ChatGPT-5
Google Launches AP2: Agent Payments Protocol for AI Commerce
Key Takeaways: Agent Payments Protocol (AP2)
Google introduces AP2, an open protocol for secure agent-led payments.
Backed by 60+ global partners, including Mastercard, American Express, PayPal, Coinbase, Salesforce, ServiceNow, Adobe, and more.
Designed to address authorization, authenticity, and accountability in AI commerce.
Uses Mandates and verifiable credentials (VCs) to establish trust.
Supports a wide spectrum of payment methods, from credit cards to stablecoins.
Includes an A2A x402 extension for crypto and Web3 payments.
Open collaboration via GitHub repository for specifications, documentation, and reference implementations.
AP2: Google’s Agent Payments Protocol for AI Commerce
Google has unveiled the Agent Payments Protocol (AP2), an open protocol developed with leading payments and technology companies to securely initiate and transact agent-led payments across platforms.
Built in collaboration with more than 60 global partners, AP2 establishes a universal framework for AI-driven commerce, supporting secure transactions across credit cards, stablecoins, real-time bank transfers, and more.
The protocol builds on the Agent2Agent (A2A) and Model Context Protocol (MCP) standards, creating a payment-agnostic foundation that ensures users, merchants, and financial institutions can transact with confidence while avoiding fragmentation in the AI agent ecosystem.
Why a Payments Protocol Is Needed
AI agents are becoming increasingly capable of conducting purchases, subscriptions, and bookings on behalf of users. While today’s payment systems assume a human clicks “buy” on a trusted interface, autonomous transactions break this assumption and raise new risks. Without safeguards, merchants may struggle to confirm intent, users may lack recourse for errors, and financial institutions may face increased exposure to fraud.
AP2 was designed to resolve these challenges by addressing three foundational requirements:
Authorization → Establishing cryptographic proof that a user explicitly granted authority for an AI agent to make a specific purchase, within defined limits. This proof works like a digital signature, creating a tamper-proof record that merchants and payment providers can verify before completing the transaction.
Authenticity → Giving merchants confidence that the agent’s request reflects a user’s true intent, reducing the risk of manipulated or fraudulent orders.
Accountability → Creating an audit trail that clarifies who is responsible if a transaction is disputed, incorrect, or fraudulent.
Deloitte underscored the urgency of these standards:
“As Agentic Commerce rapidly emerges as a transformative force, the industry will need robust standards to empower AI agents to transact payments securely and effectively. These standards must address critical areas such as security, identity, frictionless commerce, trust, and privacy, all while providing compatibility with the existing global payments infrastructure.” – Gopal Srinivasan, Alphabet Google Alliance Global AI & Data Leader at Deloitte Consulting LLP
By introducing a shared, open protocol, AP2 creates a common language for secure agent-led transactions, helping to prevent fragmentation across the AI agent ecosystem. It supports a broad spectrum of payment methods—including credit and debit cards, stablecoins, and real-time bank transfers—to deliver a consistent, scalable experience for both users and merchants. At the same time, it provides financial institutions with the clarity needed to manage risk effectively, ensuring the global payments ecosystem can evolve smoothly across platforms, industries, and geographies.
How AP2 Works: Mandates and Verifiable Credentials
At the core of AP2 are Mandates — tamper-proof, cryptographically signed digital contracts that provide verifiable proof of a user’s instructions. Each Mandate is validated by verifiable credentials (VCs) and serves as the auditable evidence for every agent-led purchase. Together, they establish the link between a user’s intent, the items in their cart, and the final payment, creating a trustworthy foundation for every transaction.
AP2 supports two main shopping modes:
Real-time purchases (human present):
Example: A user asks their agent, “Find me new white running shoes.”
The request is captured in an Intent Mandate, which records the user’s initial instructions in a secure, auditable format.
Once the AI agent assembles a cart with the selected shoes, the user explicitly approves it by signing a Cart Mandate.
This step creates a tamper-proof record of the exact items and price, ensuring that what you see is what you pay for.
Delegated tasks (human not present):
Example: A user instructs, “Buy concert tickets the moment they go on sale, but only if under $100.”
The user pre-authorizes this task with a detailed Intent Mandate, setting conditions like price limits, timing, or seller preferences.
When those conditions are met, the AI agent automatically generates and signs a Cart Mandate on the user’s behalf, without requiring further input.
Because the Intent Mandate is verifiable and pre-authorized, merchants and payment providers can trust that the resulting transaction fully reflects the user’s instructions.
In both scenarios, this chain of evidence — Intent Mandate → Cart Mandate → Payment — creates a non-repudiable audit trail. Each step is cryptographically secured and independently verifiable, ensuring that the three pillars of authorization, authenticity, and accountability are met for every transaction.
Unlocking New Commerce Experiences
The flexibility of AP2 opens the door to entirely new AI-powered shopping experiences that go beyond today’s e-commerce models:
Smarter shopping → Users can set precise purchase conditions—such as color preferences, size availability, or price thresholds—and their AI agents will continuously monitor for matches. For example, if a specific winter jacket is out of stock in green, the agent can watch for availability and execute a secure, verifiable purchase the moment it appears, even paying slightly more if that condition was pre-approved. This helps capture high-intent sales that merchants might otherwise lose.
Personalized offers → AI agents can pass along rich context from user requests to merchants, enabling more targeted offers. For instance, a shopper may request a new bicycle for an upcoming trip. The agent communicates the request—including the travel date—to the merchant’s own agent, which responds with a custom bundle: the bike, a helmet, and a travel rack at a 15% discount. This transforms a simple query into a tailored sale with higher value for both buyer and seller.
Coordinated tasks → Instead of handling each booking separately, AI agents can orchestrate multiple purchases across different providers. For example, a user planning a weekend trip might instruct: “Book me a round-trip flight and a hotel in Palm Springs for the first weekend of November, with a total budget of $700.” The agent can negotiate across airlines, hotels, and online travel agencies, then execute both bookings simultaneously and securely once the right combination is found. This coordination ensures budget alignment while reducing friction for the user.
Together, these scenarios highlight how AP2 enables commerce that is more responsive, personalized, and efficient than traditional checkout flows—unlocking opportunities for users, merchants, and financial institutions alike.
Crypto and Emerging Payments Systems: Stablecoins and Standards
Beyond traditional payments, AP2 is designed as a universal protocol supporting a wide variety of payment types, including stablecoins and other cryptocurrencies. To accelerate adoption in the Web3 ecosystem, Google partnered with Coinbase, Ethereum Foundation, MetaMask, and others to launch the A2A x402 extension, a production-ready solution enabling secure, agent-based crypto payments.
Coinbase emphasized the significance of this milestone:
“x402 and AP2 show that agent-to-agent payments aren’t just an experiment anymore — they’re becoming part of how developers actually build.” – Erik Reppel, Coinbase Developer Platform
Coinbase CEO Brian Armstrong added a broader perspective, noting:
“Agents can actually pay each other now, with x402 powering the stablecoin rail inside Google’s new Agent Payments Protocol (AP2). Really cool.”
Armstrong also showcased a demo with Lowe’s Innovation Labs, where an AI agent purchased a refrigerator using AP2 + x402, demonstrating the protocol’s potential for real-world commerce.
Traditional financial leaders are also backing this effort. Mastercard Chief Digital Officer Pablo Fourez explained:
“Mastercard is committed to ongoing, responsible innovation – and we are excited to be collaborating with Google, leading banks, merchants, AI platforms and other industry leaders to help shape the future of agentic commerce. These efforts include critical work with standards bodies such as the FIDO Alliance, where we are advancing verifiable credentials to capture and secure consumers’ intent in this dynamic new context.”
Industry Collaboration and Support
The ecosystem backing AP2 is extensive, spanning payments leaders, consultancies, and technology providers. To illustrate the breadth of support, here are perspectives from some of Google’s partners helping to shape the protocol:
Financial networks and processors:
“With the rise of AI-driven commerce, trust and accountability are more important than ever. American Express is excited to contribute to the creation of AP2 as a protocol intended to protect customers and enable participation in the next generation of digital payments.” – Luke Gebb, EVP, Amex Digital Labs, American Express
“AP2 provides the critical foundation for trusted agent payments, giving the ecosystem much needed clarity on how to facilitate trusted transactions. PayPal is fully aligned with this vision and excited to build on it, bringing our commerce expertise to help extend these principles across the entire purchase journey.” – Prakhar Mehrotra, SVP & Global Head of AI, PayPal
“Worldpay shares Google’s vision of an open, interoperable foundation for agentic commerce, built on trust and safety to empower merchants and shoppers. The AP2 protocol represents a meaningful first step in defining how agents, merchants, and payment providers can transact securely, at scale.” – Cindy Turner, Chief Product Officer, Worldpay
Technology and commerce providers:
“Adobe is proud to work with Google to advance secure and authenticated agentic commerce - our role in the Agent Payments Protocol (AP2) underscores our commitment to trusted, AI-driven experiences. With Adobe Commerce and AI agents powering customer journeys, we are focused on delivering secure, reliable, and authentic transactions for businesses and consumers.” – Loni Stark, VP of Strategy and Product, Adobe
“Salesforce is excited to help businesses harness agentic payments at scale - creating truly frictionless commerce experiences and driving the productivity that is crucial to becoming an Agentic Enterprise today.” – Nitin Mangtani, SVP & GM, Commerce & Retail Cloud, Salesforce
“Together, we’re advancing the next generation of sales and procurement workflows—rooted in trust, security, and governance—while setting a new standard for how enterprises scale with agentic AI.” – Jon Sigler, EVP & GM, AI Platform, ServiceNow
Consultancies and infrastructure:
“Innovations like AP2 will enable many of the agentic solutions that reinvent payments for clients – not only for today’s needs, but for the evolving models of future commerce.” – Scott Alfieri, Google Business Lead, Accenture
“PwC is committed to fostering innovation with agentic AI that focuses on maintaining trust, safety and privacy for critical tasks like payments and money movement broadly. We believe the Agent Payments Protocol (AP2) and extension to the Agent2Agent protocol represent a significant leap forward, enhancing safety without compromising information.” – Scott Likens, PwC
The diversity of support demonstrates that AP2 is not just a Google initiative but a global foundation for the future of agent-led payments and commerce.
What’s Next: Open Collaboration and GitHub Access
Google is calling on the global payments and technology community to actively build on AP2 through an open, collaborative process. The company emphasizes that the protocol’s success depends on broad participation, from developers experimenting with new use cases to enterprises integrating it into complex financial workflows.
To support adoption, Google has published the full technical specifications, documentation, and reference implementations in the public AP2 GitHub repository, which will be regularly updated with new examples and innovations.
Future contributions—from both Google and the wider ecosystem—are expected to showcase the scalability and interoperability of AP2 across industries. Potential applications range from consumer commerce (frictionless retail transactions, personalized offers) to B2B procurement (autonomous purchasing of partner-built solutions, automatic scaling of software licenses based on real-time demand).
By inviting the community to test, extend, and refine the protocol, Google aims to establish AP2 as the trusted backbone for the next generation of agent-driven commerce.
Q&A: Agent Payments Protocol (AP2)
Q: What is AP2?
A: An open protocol developed by Google and 60+ partners to enable secure, agent-led payments across platforms.
Q: Why is AP2 needed?
A: To address authorization, authenticity, and accountability in AI-driven commerce, where agents transact on behalf of users.
Q: How does AP2 work?
A: Through Mandates and verifiable credentials, creating an auditable trail from intent to cart to payment.
Q: Does AP2 support crypto?
A: Yes. With partners like Coinbase and MetaMask, AP2 includes the A2A x402 extension for crypto and stablecoin payments.
Q: How can developers get started?
A: By accessing the AP2 GitHub repository for specifications, documentation, and reference implementations.
What This Means: A New Era for AI-Driven Commerce
The launch of AP2 represents a pivotal step in standardizing AI-driven payments. By creating a shared foundation, Google and its partners are ensuring that AI agents can transact securely across ecosystems, bridging traditional finance and emerging payment systems.
It also marks a turning point in the evolution of AI agents. For the first time, autonomous systems can securely conduct real financial transactions on behalf of people and organizations. This shift is not incremental—it represents a foundational change in how commerce will function.
For consumers, this means everyday tasks like shopping, booking travel, or managing subscriptions can be delegated to trusted AI agents, saving time and reducing friction.
For merchants, it unlocks higher conversion rates and new revenue streams, powered by personalized offers, high-intent purchases, and seamless coordination across platforms.
For financial institutions, it establishes a standardized framework to manage risk while participating in an increasingly agent-driven economy.
The implications are profound. By empowering AI agents to act as economic participants, AP2 redefines the relationship between people, commerce, and technology. What was once science fiction—autonomous digital agents making secure purchases—is now becoming reality.
The message is clear: AI is no longer just a tool for generating information or insights. With AP2, agentic commerce has the infrastructure to scale safely and globally, stepping directly into the flow of money, commerce, and trust. This transformation brings enormous opportunities but also underscores the need for careful standards, strong safeguards, and open collaboration across industries.
If implemented responsibly, this new foundation could make commerce more accessible, efficient, and personalized than ever before—reshaping the digital economy for decades to come.
Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.